Kioxia (285A) is in turmoil.
On June 22, it hit an all-time high of 112,700 yen. In just over three weeks, by July 16, it dropped to the 62,000 yen range.
A decline of about 44%.
Moreover, earnings didn't deteriorate during this period. Last fiscal year, sales were 2.3376 trillion yen (+37%), and operating profit was 870.4 billion yen (+93%). The gross margin was a staggering 84.9%.
In the midst of record-high profits, only the stock price has been slashed by nearly half.
"Why?"
It's natural to wonder.
① Why did it fall?
There are three triggers.
Trigger 1: The question posed by Meta
The catalyst was concern over Meta's AI capital expenditures.
"Can this investment really be justified?"
The numbers didn't worsen. However, for stocks trading at high multiples based on the premise that "AI demand is infinite," the mere appearance of a crack becomes a reason to sell.
Kioxia specializes in NAND. In other words, it is the stock most purely linked to "AI storage demand." Therefore, the moment skepticism about AI demand emerged, it reacted more sharply than any other semiconductor stock.
Trigger 2: South Korea's 84 trillion yen production increase plan
Samsung and SK Hynix announced plans for capital investments on a different scale. If supply increases, prices will reverse. This is the destiny of the memory industry.
Since Kioxia's performance is directly linked to NAND prices, the market began to anticipate a "price collapse in a few years."
Trigger 3: The 84.9% gross margin figure
Here is the scariest lesson in stocks.
An 84.9% gross margin is proof of strength, but at the same time, it suggests there is almost no room for further improvement. Historically, memory stocks have peaked when their performance was at its best.
"Peak performance = Stock price ceiling."
Institutional investors remember this pattern all too well. So, the moment the peak is in sight, they start offloading their positions.
② Who was buying and driving it up in the first place?
Without understanding this, you can't see the essence of the decline. Kioxia went public in December 2024. In 2025, it spent a long time moving modestly in the 1,000–2,000 yen range.
The engine started in 2026. At the beginning of the year, it was in the 20,000–30,000 yen range. From there, it surged to the 110,000 yen range in six months. About 4 to 5 times. In six months.
So, who bought this?
Phase 1 (Jan–Mar): Institutional Investors and Foreigners
A sense of undervaluation after listing, a rapid recovery in earnings, and the full-scale start of AI demand. Overseas hedge funds and institutional investors who noticed this first bought in. The so-called "smart money." The stock price was still in the 20,000–40,000 yen range.
Phase 2 (Apr–May): Domestic Institutions and Some Individuals
Riding the wave of the semiconductor and AI boom, domestic securities firms, investment trusts, and report recommendations spread. The stock price moved to 50,000–80,000 yen.
Phase 3 (June): Late-coming Individual Investors, specifically using margin trading
This is the foreshadowing of the current decline. As the stock price exceeded 100,000 yen, voices saying "it will still go up" spread on SNS and YouTube. Individual investors bought in using margin trading.
As of early July, the margin buying balance surged, and the margin ratio exceeded 27 times. This is an abnormal figure where "there are only people who want to buy and almost no one who wants to sell." In other words, the final push for the rise was "margin buying by individuals chasing high prices."
This is a very important point.
③ Who is selling now?
The structure becomes clear when you track the actors chronologically.
First to sell was the smart money
From late June to early July, skepticism about AI capital investment spread. The first to catch this were the overseas institutions that had bought in the early stages of the rise. They had already gained 4 to 5 times returns. They started taking profits based on a "vague atmosphere." This was the first wave of selling.
Next came the forced selling by individuals on margin
When the stock price falls, unrealized losses on margin trades swell. To avoid margin calls (additional collateral), individuals are forced to cut their losses. This triggered a typical chain reaction of selling.
A 14.89% drop on July 2, and a 14.92% drop on July 16.
Violent drops like these don't happen solely based on institutional judgment. They occur when individual margin traders are forced into a state where they have no choice but to sell.
And now, short sellers are dealing the final blow
Stocks with broken supply-demand dynamics become targets for short selling. If hedge funds think "it will fall further," they will sell relentlessly. The crash from the 70,000 yen range to the 60,000 yen range is the result of these three layers of selling occurring simultaneously.
Importantly, as of July 13, there are still over 13 million shares in the margin buying balance. There is still a backlog of inventory waiting to be dumped. Until this is cleared, the upside remains heavy from a supply-demand perspective.
④ What will happen in the next 6 months to a year?
This is likely the biggest concern.
Short-term (~End of July): July 31 earnings are the first hurdle
The numbers themselves are likely to be good. Given the gross margin of the previous period and the upward trend in NAND prices, it's hard to produce bad results.
However, what the market looks at is the "forecast for the current period." If the company can issue a bullish outlook, the decline will stop. If they issue a cautious outlook or hint at a "NAND price peak-out," there is a possibility of further decline.
Since margin buying balances remain, an additional 10–20% drop is possible in the case of a downward surprise.
Mid-term (Autumn–Year-end): The real deal is US Mega-cap earnings
What really moves Kioxia's stock price is actually not Kioxia's own earnings. It's the timing when Meta, Microsoft, and Google declare "how much AI capital investment they will do in 2027."
Earnings are from August to October. If they say "we will significantly expand capital investment in 2027 as well," the reasons for Kioxia's decline will mostly disappear. Conversely, if they say "we're taking a breather," the entire AI market will enter a correction.
Kioxia is on the front lines. In other words, Kioxia's stock price from autumn onwards depends not on Japanese factors, but on the earnings of US Mega-caps.
Long-term (1 year later~): Three turning points
Looking long-term, there are three branches for Kioxia.
Branch A: If AI demand is real, it's a major buying opportunity
NAND demand is likely to continue expanding through the twin engines of Generative AI and Edge AI. Micron management has stated that "supply and demand will remain tight beyond 2027." If this is correct, the current 60,000 yen range will be an exceptional buying opportunity.
Branch B: If prices reverse due to Korean production increases, performance will also peak out
NAND prices could reverse around 2027 when the production increases by Samsung and SK Hynix actually begin. Kioxia is losing in terms of scale. If that happens, both performance and stock price will confirm a peak-out.
Branch C: Kioxia's restructuring
Personally, I think this is the biggest story. Kioxia is not of a size that can compete alone against the scale of the South Korean players. Rumors of a management integration with US-based SanDisk have been smoldering for years.
The fact that the market capitalization has decreased due to this crash also acts in a direction that increases the possibility of integration talks moving forward. If restructuring actually happens, the stock price will show another level of volatility. This is a long-term theme for the next year or several years.
Summary (What to look for in this stock)
Kioxia's movement is not just a "crash of a NAND stock." For Japanese investors, it is a stock that functions as a thermometer for the entire AI market.
If you look at "why it fell" in isolation, it looks like deteriorating supply and demand.
If you look at "who bought and who sold," you can see it's a story of the collapse of individual margin buying that supported the end of the rise.
If you look at "what will happen next," an unexpected structure emerges where it actually depends on the earnings of Meta, Microsoft, and Google.
Regardless of whether you trade Kioxia itself, just following the price movement of this stock allows you to read the tide of the entire AI market.
In that sense, I believe it is the "stock most worth watching" in the Japanese market right now.
The earnings on July 31, and the subsequent trends of US Mega-caps.
I won't be able to take my eyes off it for a while.
Note: This article is not based on interviews or supervision by Mr. BNF himself. It is the author's analysis based on public information. It does not recommend specific stocks. It is all virtual soliloquy. Invest at your own risk.
